Understanding Well-being Data

chapter 4 Discovering ‘the new science of happiness’ and subjective well-being

What well-being means to people is subjective

While we have covered what subjective well-being means previously, it is important to note that what well-being means for people in their everyday lives is subjective. Recalling the free text field analysis discussed in Chap. 2, when people are asked what is important to their well-being, they present different kinds of answers, about different areas of their life.[1] Similarly, you might look at the aforementioned four questions from the ONS and think, ‘well they don’t capture my well-being!’ You might also think about how your answer to a question about life satisfaction will have fluctuated across a year, or even a day: meanings may not be constant and bad days at work or a bad commute will make it fluctuate, affecting how you might answer the questions on how satisfied and happy you are overall. Alongside these smaller, more everyday interferences to our mood are the major events, such as grief, injury, sudden or long-term unemployment, divorce, or of course, the generalised anxiety caused by an international pandemic. Answers to these questions can reflect a fleeting positive experience, such as attending a concert, or reflect something you are missing out on, on a longer term: good relationships, a stable job, mobility or good mental health. When we come to the different measures, we shall see how these are accounted for—to a degree.

As we shall discover, subjective well-being is complex to capture in a way that can inform behaviour. There are often trade-offs to supposedly positive choices. People who enter into adult education as mature students, for example, gain the pleasure of learning and feeling purpose in their life [2], and although the negative effects are less studied [3], people miss many hedonic aspects of subjective well-being that they were previously used to, because time and energy for social and leisure activities are further compromised [4]. The same can be seen in data about parenthood [5]: it’s rewarding, but you lose fun, time, money and autonomy; other relationships suffer and it can be unexpectedly lonely [6]. A simpler binary, as found by White and Dolan [7], is that time spent with children is relatively more rewarding than pleasurable, whereas time spent watching television is relatively more pleasurable than rewarding.

The measurement of well-being aims to capture how life is lived in society so that we can know how people are getting on. But this happens at a scale that means the subjective experience of well-being can be lost. Different people have different opinions on whether this is important to the overall measurements of well-being of populations. Experts who are great with numbers work on the basis that if your unit of analysis is a population (as in population level), and as long as those whose experiences don’t fit the story are outliers, then, it will statistically even out. Therefore, crucially, these measures are not necessarily meant to capture how everyone feels about everything. Instead, they are meant to be able to compare whether particular groups are affected or how things might change over time. The aim of these measures is to do better at measuring how people are doing overall, so that better policy decisions can be made.

Others argue that measuring well-being can obscure ill-being,[8] particularly in already marginalised populations [9]. There is concern that people who are already vulnerable are placed at further risk through the way that policy deals with data. For example, an issue which has gained prominence since the #MeToo movement is sexual harassment in universities. These cases can be obscured as they might be considered ‘outliers’, and so not get picked up by data which looks for overall well-being trends [10]. Similarly, marginalised experiences of ill-being are generally less visible [11]. In Chap. 3, we briefly touched on the capacity of the domains and indicators in the OECD index, and how unlikely they would be to find the impact of policy change, like Bogue’s research on the ‘bedroom tax’. Capturing well-being data at scale, therefore, does not always pick up the complexity or subjectivities of ill-being.

The second wave of well-being is distinguished from the first, because it sees the collection of data about how people feel, at scale. For this to be effective, people need to relate to the ideas of well-being they are being asked to think about in the survey questions used. However, people do not always relate to the task at hand, or, even understand the questions asked. In my primary research, people talked about how they felt about the idea of measuring well-being [12], as they did in the ONS’ national consultations [13]. In both cases, some said it was a waste of time; that we have more important things to worry about. Others said that they didn’t understand how what is measured reflects their experience, or they didn’t understand the questions [14]. As we will discover, the ONS also found this when they trialled the ONS4. So, although subjective well-being measures are thought more democratic (because they are about how people feel), they are—of course—by and large decided by experts and defined by experts, who preside on advisory boards and write influential working papers to the ONS and international agencies. What we see is a tension between ‘robust approaches’ and ‘understandable to everybody’.


  1. In Oman 2015a , where I discuss my reanalysis of the UK’s Measuring National Well-being debate, I present the complex, heart-breaking and rich narrative of a specialist nurse, who had become unemployed owing to her own ill-health (pp. 81–82). This might be compared with more expedient free text answers of only a few words[]
  2. Duckworth and Cara 2012[]
  3. Field 2009[]
  4. Aldridge and Lavender 2000[]
  5. i.e. Pollmann-Schult 2014[]
  6. Oman and Edwards 2020[]
  7. 2009[]
  8. Ill-being, as you might expect, describes poor well-being, or to be more exact a deficiency in well-being.[]
  9. Ahmed 2012; Tate 2016, 2017[]
  10. Oman and Bull 2021, forthcoming[]
  11. Tate 2016; Oman and Bull 2021, forthcoming; Oman et al. 2015[]
  12. Oman 2017a[]
  13. as discussed in Oman 2015a, 2020[]
  14. Oman 2015a[]